Caculate Anti Asian Rate over all kinds of hate crimes for each year:

Select the mean of 2009-2018 Anti Asian Rate in each state as our Model:

Select 2019 Anti Asian Rate in each state as our Observation:

Combine them to data frame df4.

Caculate P(O|M):

$P(O|M)\approx1-|O-E|$

Caculate P(M|O):

$p(θ|y)=p(y|θ)p(θ)$

Caculate $Surprise=D_{KL}(P(M|O)||P(M))$:

image-2.png

USA 2019 Asian Hate Crimes Rate Observed Heat Map

USA 2019 Asian Hate Crimes Bayesian Surprise Map based on last 10 years

Different obejctive: USA 2019 Asian Hate Crimes Bayesian Surprise Map based on last 20 years

Comparing 3 maps